Tips for writing up a phonological analysis

(inspired by handouts distributed at UMass, UCSC, and Rutgers)

Here are some important strategies for producing a good phonological analysis write-up. It's a good idea to become familiar with them, and to check your own work against these criteria before turning it in.

Overall structure: Organization and justification of your analysis

• Avoid the "history of the thought process" structure. Begin your write-up by briefly stating your proposal. The rest of your write-up is a justification of this proposal, and (where applicable) a formal implementation.

• Remember "description before analysis." Each argument you make begins with a phonologically insightful prose description of the phenomenon or pattern to be accounted for. Only after a clear description has been presented does your formal analysis begin.

• Any claims or generalizations you make about patterns in the data are explicitly supported with well-organized examples from the data set.
  • Do not begin your write-up with a single giant data list (unless you have been specifically requested to do this). Instead, introduce particular sets of examples at the point where they become relevant to your discussion.
  • It is better to support an argument with several examples from the data set than to give just one. In a small data set, you might give all relevant examples.
  • You may wish to present examples of cases that are not part of a pattern, to show that your characterization of the pattern is sufficiently restrictive.
  • In nearly all cases you should give your data in a list or table format, set off with an example number. This is much more helpful to your reader than embedding them in a text paragraph.
  • Never assume that your reader is looking at the problem set assignment sheet. Make your write-up a self-contained linguistics paper, if a small one.

Characteristics of a good phonological analysis

• In identifying natural classes, morphological factors are considered only as a last resort, when reference to phonological environments alone is truly insufficient (and you have explicitly demonstrated this).

• Your analysis correctly accounts for virtually all of the data. (Significant over- or under-generation of the data is a serious problem.)

• Rules, constraints, or other phonological formalisms are correctly and appropriately employed. In addition, formal notation is paraphrased in prose form.
  • Correct use of formal notation is important. A global goal in phonological theory is to to build and refine a formal model that captures language-particular and universal aspects of phonological patterns. To evaluate our model, we need to be able to understand and demonstrate what it can and cannot do.
  • However, it is also important to state in prose what the formal notation is intended to accomplish. Formalisms can go out of date and be hard for later generations of linguists to read. Also, you may have made an error in your notation, so the prose statement is a backup device that tells the reader what the formal analysis is intended to do.

• In a rule-based analysis:
  1. The number of rules required for the analysis is the minimum reasonably possible; generalizations are captured.
  2. Crucial rule orderings are demonstrated by (mini-)derivations comparing good and bad orderings.
  3. There are full summary derivations for a few representative URs that are correct and consistent with the proposed rule formulations.

• In an OT analysis:
  1. Constraints are explicitly defined, using the appropriate formalism for phonological representations. It must be unambiguous to the reader under what conditions the constraint is violated.
  2. Crucial constraint rankings are demonstrated by (mini-)tableaus that give legitimate ranking arguments. "Comparative tableaus" are particularly useful for this.
  3. Sufficient numbers of relevant losing candidates are considered so that as many constraints as possible can be crucially ranked.
  4. A final ranking summary is given that shows all and only the ranking relationships that can be explicitly proven among the constraints you have considered.
  5. There are summary tableaus for a few representative inputs that include all the constraints in the analysis. They are correct and consistent with the patterns observed in the language.

• When appropriate, alternative analyses are considered and rejected on reasonable grounds.

• The discussion is clear, consistent and complete. Terminology is correctly used.
  • Your linguist audience will be more impressed by well-organized thinking and clarity of discussion than by flowery language. An elegant writing style can sometimes be a nice thing, but not if it comes at the expense of clarity.